A plenoptic camera, also called light-field camera, acquires different views of a same scene. To obtain the different views, a raw image (comprising the data collected by a photosensor array of the plenoptic camera) is demosaiced and de-multiplexed. The demosaicing enables to recover a full color raw image, i.e. to recover full color information (for example RGB information, RGB standing for “Red”, “Green” and “Blue”) for the pixels of the raw image while the raw image acquired with the plenoptic camera associates only one color component (R, G or B for example) with each pixel. The demultiplexing performed after the demosaicing enables to recover the different views of the scene, i.e. to group the pixels of the demosaiced raw image according to the view they belong to.
To recover the different views of the scene, it is necessary to calibrate the raw image acquired with the plenoptic camera. The main goal of this calibration is to identify center point locations of each microlens micro-image formed on the photosensor array of the plenoptic camera. Each time the main lens has undergone a change, either in zooming or in focalization, the micro-image centers have to be estimated again. This creates an inconvenience because it is not very handy to calibrate the system each time the main lens has been tweaked for a snapshot. If the system is also video capable, this means also that zooming while filming is prohibited.